Building your own ComputerEase data pipeline: a DIY guide
The architecture, the stages and where the complexity sits, so your team can decide whether to build it in-house.
Last updated October 2026
Is DIY right for you?
DIY works when you already have the people. You need:
- Someone who writes SQL and can work out a table they've never seen before.
- Someone who builds Power BI models, not just reports on top of someone else's model.
- Admin access to your ComputerEase server, because the extract runs there.
- A Microsoft 365 or Azure admin who can create resources and grant access in your tenant.
- Time set aside for the project, outside month-end close.
Your team doesn't need ComputerEase data experience to start. Learning how ComputerEase stores things is most of the work, and it's where outside guidance saves the most time.
The architecture
A ComputerEase pipeline has four layers. Data only moves one way, from ComputerEase toward the people who use it.
- Source. ComputerEase on its server. The pipeline only reads from it and never writes back.
- Landing zone. A separate place where nightly files arrive and get checked before anything trusts them.
- Database. Microsoft Fabric or Azure SQL in your tenant, holding the loaded tables, their history and readable views on top.
- Consumers. Power BI, AI agents, alerts and anything else that reads the views, each with access to only what it needs.
Keeping the layers separate is what makes the system safe to change. A bad file stops at the landing zone, a schema change stops at the views, and a report never touches ComputerEase.
The stages
| Stage | What it involves | Where the complexity is | Effort |
|---|---|---|---|
| Discovery and access | Inventory the modules and tables you use, get server and tenant access, decide what stays behind | Permissions inside ComputerEase can hide data without any error | Medium |
| Extract | A scheduled job on the ComputerEase server that reads the tables you chose | The ODBC driver is old and 32-bit, and table design dates back decades | Medium |
| Land and validate | Secure upload, integrity checks, then loading only files that pass | Deciding what counts as a good file, and what happens when one isn't | Medium |
| Model | History tables and readable views that translate ComputerEase codes into plain names | Job cost, inventory and the GL each follow their own rules | High |
| Reconcile | Tie key numbers back to ComputerEase screens before anyone relies on them | Explaining every difference, including the ones that are correct | High |
| Report | A Power BI model, dashboards, security roles and a sharing plan | Measures that match how your business already talks about margin and cost | Medium |
| Operate | Monitoring, failure alerts, and fixes after ComputerEase updates | Noticing a quiet failure before someone makes a decision on stale data | Ongoing |
Most teams underestimate modeling and reconciliation. Getting the data out takes days. Getting numbers the controller agrees with takes longer.
Design decisions you'll make
- Fabric or Azure SQL. Fabric pairs closely with Power BI. Azure SQL is familiar to anyone who knows SQL Server. Both work.
- How fresh the data needs to be. Nightly covers nearly all job cost and financial reporting and keeps load off ComputerEase during the day.
- Mirror or keep history. ComputerEase purges some old records. Decide whether your copy keeps them.
- Security boundaries. Decide who sees owner-level financials and who sees operational data, and keep them apart from the start.
- What never leaves ComputerEase. Payroll, Social Security numbers, tax IDs and bank details are easiest to protect when you never extract them.
Where DIY projects stall
- Totals look right on a few jobs and wrong on others, and nobody can say why.
- Inventory and the GL won't tie, and the team can't tell a data problem from a real one.
- Reports get built on raw tables, so every report repeats the same fixes in a different way.
- The extract fails one night and nobody notices for a week.
- The one person who built it moves on, and nothing is written down.
For more on the ComputerEase side, see ComputerEase integration options and ComputerEase to Power BI.
Getting help
We offer DIY support for teams that have the talent and want an experienced second opinion: training sessions, reviews of your architecture and work, and time on call when your team is stuck. It's billed hourly at a rate we set on your fit call. DIY support on the pricing page.
If you'd rather hand off the whole build, our packages cover it at a fixed price.
Building it yourself?
Tell us about your team on a free 30-minute fit call.